Salvato in:
Dettagli Bibliografici
Autori principali: Spagnoletti, Alessio, Wang, Tim Y. J., Pereyra, Marcelo, Akyildiz, O. Deniz
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:https://arxiv.org/abs/2605.07907
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866915993877479424
author Spagnoletti, Alessio
Wang, Tim Y. J.
Pereyra, Marcelo
Akyildiz, O. Deniz
author_facet Spagnoletti, Alessio
Wang, Tim Y. J.
Pereyra, Marcelo
Akyildiz, O. Deniz
contents Vision-Language Latent Diffusion Models (LDMs) (Rombach et al., 2022) provide powerful generative priors for inverse problems. However, existing LDM-based inverse solvers typically require a large number of neural function evaluations (NFEs) and backpropagation through large pretrained components, leading to substantial computational costs and, in some cases, degraded reconstruction quality. We propose a unified Euclidean-Wasserstein-2 gradient-flow framework that jointly performs posterior sampling and prompt optimization in the latent space through a single flow that aligns the prior and posterior with the observed data. Combined with few-step latent text-to-image models, this formulation enables low-NFE inference without backpropagation through autoencoders. Experiments across several canonical imaging inverse problems show that our method achieves state-of-the-art performance with significantly reduced computational cost.
format Preprint
id arxiv_https___arxiv_org_abs_2605_07907
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Consistency Regularised Gradient Flows for Inverse Problems
Spagnoletti, Alessio
Wang, Tim Y. J.
Pereyra, Marcelo
Akyildiz, O. Deniz
Machine Learning
Computer Vision and Pattern Recognition
Vision-Language Latent Diffusion Models (LDMs) (Rombach et al., 2022) provide powerful generative priors for inverse problems. However, existing LDM-based inverse solvers typically require a large number of neural function evaluations (NFEs) and backpropagation through large pretrained components, leading to substantial computational costs and, in some cases, degraded reconstruction quality. We propose a unified Euclidean-Wasserstein-2 gradient-flow framework that jointly performs posterior sampling and prompt optimization in the latent space through a single flow that aligns the prior and posterior with the observed data. Combined with few-step latent text-to-image models, this formulation enables low-NFE inference without backpropagation through autoencoders. Experiments across several canonical imaging inverse problems show that our method achieves state-of-the-art performance with significantly reduced computational cost.
title Consistency Regularised Gradient Flows for Inverse Problems
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.07907